Unleashing power of data for early stage startups

Data is the new oil is a cliche all of us have heard. All founders recognize that they need to be data-driven and want to leverage data and analytics to run their businesses. Yet we see founders unable to use data to its full potential. Why?

Here's a few common reasons we see early-stage companies unable to leverage data effectively.?

1. Knowing what to measure: Digital?native companies generate tons of data. Every transaction, every click creates a data point. It is usually very easy to get lost in details and miss the important metrics that generate a signal?amidst the noise e.g., when looking at your website's performance there a 50+ metrics you?can review but only 3-4 really matter in the larger scheme of things.?
2. Not enough granularity: Founders who do look at data, often end up either looking at things in aggregate or do not segment it in a way that is relevant to the business e.g. looking at just daily company level revenue is not sufficient, esp you sell through multiple channels. You'll easily miss what is driving the revenue upwards or downwards.?
3. Unclear objectives: Usually companies aren't clear what they are?solving for. Various metrics become important at different times during a company's journey and are extremely contextual e.g. in early stages when you are trying to build following via social media with limited resources?organic engagement is key, as the company scales and paid media becomes central metrics like CAC, RoAS will take centerstage
4. Insufficient infrastructure: While well-intentioned founders try to be data-driven, ineffective data infrastructure e.g. no set dashboards, no regular cadence, no one custodian of data within the org, makes it difficult to timely review data.?
5. Data preparation: Another common issue we see companies deal with, esp at early-stages where data is often very manual, is avoiding granularity and segmentation because it is cumbersome to clean up the base data for it to be useful. Companies in such cases, often stick to high level numbers which take lower effort to maintain and also make it look like the company is being data-driven.?
6. Lack of data culture: Often the MIS person or the analytics team becomes the owner of data and metrics, with blank stares from function owners during review meetings when the data throws up something unexpected. Business needs to own each?metrics and know why something is off when it is. The analytics team is only an?enabler.?
7. Not understanding context: Any number in isolation means nothing. It?needs to be held against a?benchmark, plan, or trajectory. e.g. a page?load time of 2 secs in isolation may not seem bad, but if your peers operate at 0.4 secs, you are 400% slower.?
8. Confirmation bias: Founders often take away from data, which reinforces their already existing intuition and belief. As a result, counter-intuitive?yet powerful insights get ignored as "aberrations in data".?

Here's how we look at data-driven decision?making esp at early-stages:

1. Bridging gap between data & value: We believe in focussing on 4-5 key metrics to track business health but tracking many more (15+). Only 4-5 metrics are going to drive most value creation in your business, the rest are tracked for when you need to?dig deeper because one of your main metrics changed dramatically
2. DIstribution vs averages: Averages hide the real insights. We look at distributions and sub-segments to find the true driver of a trend.?
3. Trend matters as much as data: It is important to often look at directional numbers rather than specific,?e.g. when deciding what new products to launch based on gross margins, GM% may not be accurately available down to the second decimal, but just relative prioritization?is also good enough?
4. Inputs vs outputs: Output can take a long time to reflect the impact of your actions, so tracking inputs is equally important. e.g., if you are looking to?improve RoAS (Return on Ad Spend) how many ad copies are you testing each week to find that winning creative is an important metric given RoAS improvement can be a more long-term process
5. Defining good enough data: Often the more real-time data is the less accurate it maybe. It is a trade-off between speed and accuracy. Sometimes data that is?+/-5% accurate still provides enough accuracy to make directional business decisions.?
6. Defining clear ownership: Each metric, channel, etc. needs to have a ONE clear owner. Assigning it to a group just obfuscates decision making and slows down action
7. Top-down checks: Before you take a certain metric are face-value, important to make quick top-down sense checks to make sure?nothing is obviously off, e.g. when reviewing channel-wise spends?based on contribution margin, first check if the margin for individual channels adds up to the overall
8. Discipline of regular tracking: We encourage our companies to track data daily even if some of those metrics get reviewed weekly/monthly. A daily cadence builts habits and allows data-driven micro decisions, which compound in the long-run.?

We would love to hear your thoughts how do you leverage data in your day to day decision making.